Skip to main content
SearchLoginLogin or Signup

The Comparative Influence of Majority and Expert Opinion (1921)

1921. American Journal of Psychology 32: pp. 16–20 [with elisions].

Published onApr 09, 2024
The Comparative Influence of Majority and Expert Opinion (1921)


Editor’s Introduction

Patrick Parsons

Among the more heavily studied sub-topics in suggestion theory was the power of the authoritative speaker (prestige) and the reference group itself. An experiment by psychologist Henry T. Moore (1903–1967) in 1921 is illustrative.

Moore had degrees from the University of Missouri (AB, 1903), Yale (AM, 1907) and Harvard (PhD, 1914). He was on the psychology faculty at the University of Minnesota and then Dartmouth until 1925, when he accepted a position as the second president of Skidmore College in New York, where he served for 32 years.

Although Allport was a—maybe the—leading exponent of laboratory experimentation in social psychology, he held no early monopoly on the method. Moore’s work represents one of many examples of laboratory research applied to suggestibility in the 1910s and 1920s. Instead of sweeping sociological claims of the kind characterized by writers such as Trotter or Ross, Moore focused on one narrow question, using a rigorous and elaborate design to gather quantifiable information. The question at hand was “the influence of the group upon the opinions of the individual,” referencing, although not altogether favorably, Trotter’s Instincts of the Herd. Moore also sought to compare the influence of the group with the influence of an outside expert.

His design marshaled ninety-five subjects, testing responses to treatments across three topics: linguistic judgments, ethical judgments, and musical preferences. He concluded, using terms of statistical probability, that there was evidence of suggestive influence in all conditions, although it was greater in matters of speech and morals than in music. The influence of group opinion was somewhat more powerful than that of the expert in matters of speech. The expert could exert some influence on musical taste, however, so “classical music has at least an even chance in its struggle with the popularity of the jazz band.”

Referencing what we now label “generalizability,” he noted in conclusion that similar tests of suggestibility in areas of fashion or orthodoxy “would give material which would be valid for general purposes.”

The Comparative Influence of Majority and Expert Opinion (1921)

Henry T. Moore

 American Journal of Psychology 32: pp. 16–20 [with elisions].

The literature on Social Psychology contains numerous references to the influence of the group on the opinions of the individual. This one point has been made the subject of practically an entire volume by Trotter,1 who refers to group opinion somewhat picturesquely as the voice of the herd. He represents this voice as coming with such a weight of authority that even the most eccentric individual feels compelled to seek some form of herd support for his opinions, and is completely at a loss when no such support is anywhere to be had.

The general fact is beyond dispute, but those who would like to see Social Psychology multiply its experimental findings are tempted to ask more specifically just how great this influence may be expected to be in any given situation. Can we hope to measure it? And if so, how does it compare with other influences that are likely to operate in determining an individual’s social attitudes?

The group experiment here reported attempts a beginning at answering these questions for three types of situation,­ namely, speech, morals, and music. The method is somewhat similar to one used by Bridges2 in a study of decision types reported in 1914. In general it consists of measuring a suggestive influence in terms of the number of reversals of judgment occasioned by it, as compared with the number that might have been expected by chance. The first problem was therefore to find out what was the chance of reversal of judgment in regard to each of the three kinds of material used. Ninety-five subjects were given eighteen paired comparisons for each of the three types of situation. The instructions for the linguistic judgments were that the subjects check the more offensive one of each pair of expressions. Examples of the expressions compared are: “Everybody loves their mother.” “She sort of avoided him.” “The party that wrote that was a scholar.” “He never studies nights.” The ethical judgments involved the checking of the more offensive of two traits of character in each of the eighteen pairs. Examples of the traits compared are: disloyalty to friends; willingness to get rich by questionable financial methods; cheating on an examination; willingness to overlook a business error favorable to oneself. The musical judgments involved an expression of preference for one of two resolutions of the dominant seventh chord, played on a reed organ. Eighteen paired resolutions were played, and the preferences recorded after each.

Two days later the same three series were repeated exactly as given before, and without the introduction of any special suggestive influence to alter the original judgments. Each subject was now scored on the basis of his percentage of reversals, and the mean of the ninety-five individual scores was taken as the chance of reversal for judgments concerned with that particular type of material. The average score thus recorded as representing the chance of reversal for linguistic judgments was 13.5 per cent with .55 P. E. of the mean; for moral judgments 10.3 per cent with P. E. of .50; for musical judgments 25.1 per cent, with P. E. of .84.

As a partial check on the above figures each of the three series of judgments was tried on a different group. The subjects in the check experiment were on an average about a year older than those in the original experiment, which probably accounts for their slightly lower per cent of reversals. Forty-three subjects gave 11.4 per cent chance of reversal in linguistic judgments; 62 subjects gave 9.4 per cent reversals in ethical judgments; and 49 subjects gave 22.6 per cent reversals in musical judgments.

An interval of two and a half months was allowed between the experiment without suggestion and that in which suggestion was used. This seemed ample time to render negligible any memory effects from the preceding judgments. The experiment was now repeated as before, except for the addition of the suggestive influences. A new set of original judgments was taken, and after a two day interval the subjects were given the same series again, this time with the statement of what had been the majority preference for each pair. Great care was taken to convince them that these statements were being truthfully made, and the influence of suspicion was certainly not great. Each subject was now scored on the basis of the per cent of opportunities he had accepted to reverse his judgment so as to agree with majority opinion. The results are indicated in the three left columns of the upper chart. The average of such reversals in linguistic judgments was 62.2 per cent, with P. E. of 1.63; in ethical judgments 50.1 per cent, with P. E. of 1.69; in musical judgments 48.2 per cent, with P. E. of 1.52.

Two days later still the same comparisons were repeated, and at this time each judgment was preceded by a statement of the opinion of an expert in each field. These statements in some cases coincided with what had been given as majority opinion previously, but as often as not they were at variance with it. From these last records each individual was scored on the basis of the per cent of opportunities he had accepted for reversing his original judgment favorably to the statement of the expert. The results [...] show an average of 48 per cent reversals of language judgments, with P. E. of 1.4; 47.8 per cent reversals of ethical judgments, with P. E. of 1.6; and 46.2 per cent reversals in musical judgments, with P. E. of .9.

If now we take as our unit of measurement the per cent recorded as the chance of reversal, we find [...] that the probability of reversing favorably to the majority in matters of speech and morals is approximately five times chance; whereas in matters of musical feeling the probability is only about twice chance. By majority is meant here of course only the special type of majority provided in the experiment, but if generalization is permissible on the basis of the evidence available, we may venture the statement that a man is two and a half times as individualistic in his musical likes and dislikes as in his moral and linguistic preferences. Similarly we may conclude that expert and majority opinion hold about equal sway over the individual in morals and music, but that the chances are about ten to seven in favor of majority prestige in matters pertaining to speech. To take a concrete case, the struggle between the expert pronunciation which accents the word ‘cantonment’ on the first syllable, and the army pronunciation, which accents it on the second syllable, is likely to end with the doom of the expert, whereas classical music has at least an even chance in its struggle with the popularity of the jazz band.

The type of experiment here described can, on account of the great range of individual variations involved, be valid only if applied to a large number of cases; and inasmuch as each particular experiment measures only a very particular type of suggestibility, any generalization from a single experiment will always be questionable. But it is believed that an extension of the method to cover a large number of typical cases in which such social influences as personal prestige, fashion, orthodoxy, etc., play a part, would give material which would be valid for general purposes. Whether the shrunken prestige of a defeated political candidate or of an abdicated emperor follows any accurately describable laws, one could scarcely venture to say; but it is sufficiently obvious that until so-called social laws rest on more than the personal observations of individual writers, we shall have a great excess of laws, and only a minimum of confidence in applying them.3

No comments here
Why not start the discussion?